Scale-up Synthesis of Tesirine - Organic Process Research

Aug 2, 2018 - Over the course of four synthetic campaigns, the discovery route was developed and scaled up to provide a robust manufacturing process...
1 downloads 0 Views 689KB Size
Subscriber access provided by University of Sussex Library

Full Paper

Scale-up synthesis of tesirine Arnaud Charles Tiberghien, Christina Louisa Von Bulow, Conor Barry, Huajun Ge, Christian Noti, Florence Collet Leiris, Marc McCormick, Philip Wilson Howard, and Jeremy Stephen Parker Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.8b00205 • Publication Date (Web): 02 Aug 2018 Downloaded from http://pubs.acs.org on August 2, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

Scale-up synthesis of tesirine Arnaud C. Tiberghien*, † Christina von Bulow†, Conor Barry†, Huajun Ge§, Christian Noti∥, Florence Collet Leiris#, Marc McCormick‡, Philip W. Howard†, Jeremy S. Parker‡



Spirogen, QMB Innovation Centre, 42 New Road, E1 2AX London, U.K. §

Pharmaron, No.6, Taihe Road, BDA,Beijing, 100176, P.R.China ∥Lonza

#



AG, Rottenstrasse 6, CH - 3930 Visp, Switzerland

Novasep Ltd, 1 Rue Démocrite, 72000 Le Mans, France

Early Chemical Development, Pharmaceutical Sciences, IMED Biotech Unit, AstraZeneca, Macclesfield, UK.

ACS Paragon Plus Environment

1

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 55

TOC Graphic

O H N

N O

O

O

O

O

O

O

O

O

H N

O

O

H N N H

O

O O

N

H

O

O

O

O

O OH H

N

B N O

N O

SG3249, tesirine

A

N

H C

N

O

O

O

O

O

N

H N

O SG3199

ACS Paragon Plus Environment

2

Page 3 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

ABSTRACT

This work describes the enabling synthesis of tesirine, a pyrrolobenzodiazepine antibody drug conjugate drug-linker. Over the course of four synthetic campaigns, the discovery route was developed and scaled-up to provide a robust manufacturing process. Early intermediates were produced at kilogram scale and high purity, without chromatography. Mid stage reactions were optimized to minimize impurity formation. Late stage material was produced and purified using a small number of key high-pressure chromatography steps, ultimately resulting in a 169 g batch after 34 steps. At the time of writing, tesirine is the drug-linker component of 8 antibody-drug conjugates in multiple clinical trials, 4 of them pivotal.

KEYWORDS Pyrrolobenzodiazepine Antibody Drug Conjugate High pressure chromatography Loncastuximab tesirine Camidanlumab tesirine

ACS Paragon Plus Environment

3

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 55

INTRODUCTION

During the last 10 years, antibody-drug conjugates (ADCs) have made a welcome addition to our arsenal in the fight against cancer. In 2011, Adcetris (Seattle Genetics) was approved for CD30 positive AML, followed by Kadcyla (Genentech/Roche) in 2013 to treat HER-2 positive breast cancers. Last year, two further approvals (Mylotarg and Besponsa from Pfizer) validated the ADC approach, where a potent anticancer agent is delivered specifically to an antigen-expressing tumour target. The ADC field is now expanding rapidly, with more than 60 agents in clinical trials and many more in pre-clinical development.1-2 An antibody-drug conjugate is typically represented as a three component system: the targeting antibody, the linker, and the active drug (Figure 1a). Although it is desirable to optimize and tune each and every one of these components, the linker and the drug can be treated as a single small molecular entity: the drug-linker. In addition, antibody-drug conjugates are modular systems; the same antibody may be used to deliver different classes of drugs. Similarly, the same drug-linker may be conjugated to any appropriate tumor-targeting antibody. In 2012, tesirine (SG3249) was developed by Spirogen, as a drug-linker combining a set of desired properties: fast and straightforward conjugation to antibody cysteines by maleimide Michael addition, good solubility in aqueous/DMSO (90/10) systems, and a traceless cleavable linker system delivering the highly potent pyrrolobenzodiazepine (PBD) DNA crosslinker SG3199 (Figure 1b,c).3 4

ACS Paragon Plus Environment

4

Page 5 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

Figure 1: A) Schematic depiction of an antibody-drug conjugate; B) tesirine, a drug-linker featuring a dipeptide trigger; C) SG3199, the active drug released by tesirine cleavage in the presence of peptidases.

Antibodies conjugated to tesirine have been shown to be highly efficacious in preclinical studies.5-8 As a result of the modular nature of antibody drug conjugates, and the wide interest for efficacious drug-linkers, tesirine was licensed to a number of companies. At the time of writing, tesirine was the drug-linker component of more than 15 clinical trials, either in solid or liquid tumours (see Table 1), and there is a considerable requirement for robust processes providing tesirine on scale.

ACS Paragon Plus Environment

5

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Sponsor

Indication

Target

Phase

Medimmune/AZ

MM/AML

ASCT2

I

Medimmune/AZ

Prostate

PSMA

I

Medimmune/AZ

MM

BCMA

I

Abbvie

SCLC

DLL3

II Trinity

Abbvie

SCLC

DLL3

III Meru

Abbvie

SCLC

DLL3

III Tahoe

Abbvie

NETS

DLL3

I

Abbvie

Ovarian

DPEP3

I

Abbvie

Melanoma

DLL3

I/II

ADCT

Lymphoma

CD25

I

ADCT

AML/ALL

CD25

I

ADCT

B-NHL

CD19

I

ADCT

DLBCL

CD19

II

ADCT

B-ALL

CD19

I

Page 6 of 55

Table 1: Some examples of key tesirine antibody-drug conjugates clinical trials.

In 2016, we reported the synthesis of tesirine on discovery scale.9 Despite being modular and robust, the synthesis was not scalable without optimization. Indeed, flash chromatography was still used in the initial stages of the synthesis (where kilograms of intermediates are typically required to supply a phase I study), and excessively so in the later stages. A few key reactions

ACS Paragon Plus Environment

6

Page 7 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

suffered from low yields, and other were the source of impurities difficult to separate from the desired material.

Figure 2: Block synthesis of SG3249. Boxed intermediates were synthesized in facilities equipped for high-potency cytotoxics. The longest linear sequence is 20 steps long.

In this article we describe the work carried out over four synthetic campaigns, to optimize this synthetic route on scale. Chromatographic steps were eliminated as often as possible, and especially in the early stages of the route. Synthetic processes were simplified, and often telescoped. Individual yields were improved, and impurities levels were controlled. Analytical methods for each intermediate and final product were optimized and performance tests were conducted to ensure accurate monitoring of the reactions and determine the purity of the products. Ultimately, these optimizations resulted in a synthesis with improved robustness and manufacturability, a purity increase from an original 85% to 97%, and the production of a 169 g batch. This batch size appears relatively small, but because of the potency of SG3199, the drug

ACS Paragon Plus Environment

7

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 55

delivered by SG3249, only very small amounts are enough to achieve a pharmacological effect. To put this into context, a 169 g batch of SG3249 would be theoretically sufficient to provide more than 100 000 doses of Lonca-T at the current clinical schedule (120 µg/kg q3w x2).6, 10

RESULTS AND DISCUSSION In the early stages of the synthesis (Aromatic A-ring, proline C-ring, and Val-Ala peptide trigger), we sought to optimize the process for larger scale production. In particular, great efforts went into producing pure material whilst eliminating all chromatography steps. Nitro-aromatic acid A-ring 6 Scheme 1: Improved synthesis of TIPS-protected 6-nitrovanillic acid 6

Reagents and conditions: (a) K2CO3, NMP, BnBr, 60 °C, 93%; (b) AcOH, HNO3, 22 °C, 75%; (c) TFA, AcOH, 80 °C, 85%; (d) TIPSCl, triethylamine, THF, 10 °C, 100%; (e) Sulfamic acid, NH4OH, NaH2PO4, NaClO2, THF, water, 0 °C, 68%.

ACS Paragon Plus Environment

8

Page 9 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

Benzylvanillin 2 is commercially available, but it was found to be more economical to produce it on site from vanillin and benzyl bromide. Originally, DMF was used as the solvent, but a screen revealed NMP to be kinetically advantageous, with good conversions in 1h at 60 °C. Isolation was straightforward after precipitation in water. Nitration of benzylvanillin was initially performed at 12 °C, aiming to minimize by-product formations, and in particular, the ipsonitrodeformylation11 product 7 (Scheme 2a). This side product is explained by the position of the formyl para to the benzyloxy group. In 1996, Cotelle and Cateau12 postulated that ipsonitrodeformylation can occur if the formyl group occupies the most (or one of the most) electronrich position of the aromatic ring, thus explaining competition with nitration in position 6. Scheme 2: a) Ipso-nitrodeformylation of benzylvanillin; b) Photodecomposition of 6nitrobenzyvanillin

The nitration of benzylvanillin was found to be exothermic, and concerns over thermal accumulation on scale forced us to reassess the process. The order of addition was reversed

ACS Paragon Plus Environment

9

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 55

(adding a solution of benzylvanillin to the nitrating mixture), and the operational temperature was changed to 22 °C. At this temperature, nitration occurred almost immediately with very little thermal accumulation, and with a manageable level of 7 (12% by HPLC). An reaction calorimetry test (RC1) on 587 g of benzylvanillin (Figure 3) found heat generation to be easily controlled by the rate of addition. The heat of the process was 85.35 kJ and the corresponding molar enthalpy (∆Hr) was 147 kJ/mol. These parameters, together with the onset decomposition temperature of product 3 at 244 °C validated process safety for kg scale production. Precipitation in water followed by recrystallisation in ethyl acetate afforded the 6-nitrobenzylvanillin 3 in good purity (96.6%) and 75% yield. The product was protected from light, since onitrobenzaldehydes are known to photochemically convert to o-nitrosobenzoic acid (Scheme 2b).13 Next, the benzyl ether was cleaved with TFA at 80 °C. The volume of TFA was kept to a minimum (2 V) because of its cost and potential environmental impact. AcOH (1V) was added to ensure efficient stirring. Phenol 4 was cleanly obtained by precipitation with heptane. With an overall yield of 59% and a purity of 98.7%, the results for this 3 steps sequence were found in line with the reports of Rakshit and co-workers.14 On research scale, we had protected phenol 4 in a solvent-free reaction; effectively melting TIPS-Cl and imidazole at 100 °C. Although this approach is attractive in terms of solvent volume reduction, the process was difficult to control on scale, and the reaction was potentially reversible. Instead, a base and solvent screen showed high conversions with triethylamine and TIPS-Cl in DCM. Crude aldehyde 5 was used directly in the Pinnick oxidation to provide carboxylic acid 6. Here, the chlorine scavenger was swapped from hydrogen peroxide, to sulfamic acid, as described by Lindgren in his original conditions.15 This has many advantages including improved safety (no oxygen production, scavenging of chlorine dioxide), and clean

ACS Paragon Plus Environment

10

Page 11 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

removal in the aqueous phase during work-up. After slurrying in heptane, pure A-ring 6 was thus obtained in 40% yield over 5 steps, on kg scale, and avoiding any chromatography.

Figure 3: Reaction calorimetry test on benzylvanillin. 587 g of 2 (23.8 w/w %) was dosed into the reaction mixture at 20 °C during 2 h. The mixture was stirred for an additional 1h at the same temperature. Heat generation in pink. The sharp peak is caused by precipitation of the product.

ACS Paragon Plus Environment

11

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 55

Pyrrolidine C-ring 14 Scheme 3: Improved synthesis of Pyrrolidine C-ring 14

Reagents and conditions: (a) K2CO3, MTBE, water, Cbz-Cl, 15 °C, 100%; (b) MeOH, DCM, H2SO4, 40 °C, 81%; (c) THF, Water, LiCl, NaBH4, 17 °C, 71%; (d) Toluene, triethylamine, TBSCl, 25 °C, 100%; (e) IPA, 10% Pd/C, H2, 30 °C, 61%. The synthesis of pyrrolidine C-ring 14 was conducted in parallel. The route described by Gregson and co-workers16 was adapted on scale (Scheme 3). Benzyloxycarbonyl protection of trans-hydroxyproline 9 was achieved more advantageously in a MTBE/water system rather than toluene/water, with a high purity grade of Cbz-Cl in 100% yield. Classical esterification conditions employing methanol and catalytic sulfuric acid were found difficult to work-up on kg scale. Instead a 5/1 mixture of DCM/methanol was used, allowing for a convenient aqueous

ACS Paragon Plus Environment

12

Page 13 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

work-up with NaHCO3. Ester 11 was isolated in 81% yield (7.8 kg) after solvent evaporation. The reduction of the methyl ester to alcohol 12 had been readily achieved with LiBH4 on research scale. However, the quantities of reactive LiBH4 involved on kg scale posed a fire hazard, thus substitution with the more stable NaBH4 was desirable. Trials with NaBH4 alone showed that the reaction rate was much slower than with LiBH4. Lithium chloride was added to produce LiBH4 in situ thus conserving the initial reaction kinetics and low number of equivalents.17 Next, achieving the selective silylation of a primary alcohol in the presence of secondary alcohol 12 was found to be key to avoiding chromatography on scale. Historically, this improvement of selectivity had been achieved by replacing imidazole with the bulkier DBU, and a low number of equivalents of TBS-Cl (0.77 equiv); altogether with moderate success. Chromatography was still required to remove the bis-silylated product, and recycle the starting material. Here, further screenings revealed triethylamine to be a more appropriate base for this transformation.18 The rate of the reaction was reduced, but higher selectivity ratios were obtained (i.e.: SM/Primary silylation/Bis-silylation 6/89/5). The chromatography stage was eliminated and the crude 13 was used directly in the next step. After hydrogenolysis of benzyl carbamate 13, the resulting amine was purified by precipitation as its oxalate salt. The impurities were removed in the filtrate, and amine 14 was isolated as its free base in 61% yield over two steps after a basic aqueous work-up. Altogether, 14 was synthesized in 35% over 5 steps, with high purity (99.5%), on kg scale without chromatographic purification. It was later established that high purity was necessary, both for A-ring 6 and C-ring 14, to achieve good yields in the subsequent amide coupling.

ACS Paragon Plus Environment

13

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 55

Branching A and C ring aniline 19 Scheme 4: Improved synthesis of key intermediate 19

Reagents and conditions: (a) EDCI, HOPO, DCM, 15 °C, 82%; (b) TEMPO, KBr, NaOCl, NaHCO3,

3

°C,

92%;

(c)

Tf2O,

2,6-lutidine,

toluene,

-40

°C,

(d)

MeB(OH)2,

Pd(dppf)Cl2.CH2Cl2, K3PO4, toluene, 65 °C, 44% (two steps); (e) Zn, AcOH, EtOH, water, 5 °C, 81%.

ACS Paragon Plus Environment

14

Page 15 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

The conditions for the coupling of 6 and 14 reported in the research paper9 (Scheme 4) suffered from a number of issues; DCC could not be removed easily by work-up, and HOBt is classed as a desensitized explosive, causing shipping restrictions and supply chain concerns. Instead, EDCI and HOPO19 were used. Experiments conducted without HOBt or HOPO were substantially lower yielding, thus demonstrating the activating properties of these agents. The high purity of the starting materials, together with careful temperature control meant that very little impurities were produced during the reaction. The coupled product 15 could be isolated in 82% yield and 99.8% purity after crystallization from ethanol/water, on a 5 kg scale. In the next step, the secondary alcohol was oxidized to a ketone with TCCA/TEMPO. This combination is very efficient, but can result in undesired chlorination of aromatic groups and alkenes. We successfully substituted TCCA/TEMPO with DMP (89% on 4 kg batches). However, DMP is costly, shock sensitive and the work-up can be challenging. With this in mind, the team selected the simpler and cleaner Anelli-Montanari process20 (TEMPO/Bleach in buffered biphasic conditions). Ketone 16 was obtained cleanly in 92% yield. Next, ketone 16 was transformed to the thermodynamic enol triflate 17 (with the unsaturation conjugated with the nitrogen in position 2,3). Although the conditions of this reaction were not dramatically altered, subtle changes in the number of equivalents of triflic anhydride (1.5 equiv instead of 3 equiv), and 2,6lutidine (2 equiv instead of 4 equiv) meant that the amount of 2,6-lutidine triflate by-product was considerably reduced. This by-product acted as a poison in the subsequent Suzuki coupling and had to be controlled to a low level. The solvent was switched from DCM to toluene, allowing telescoping of the triflation with the next step. Introduction of the methyl group in C2 by sp2-sp3 Suzuki coupling was particularly challenging. The operational temperature of 65 °C is very close to the decomposition temperature of the triflate. Additionally, undesired reduction of 17 resulted

ACS Paragon Plus Environment

15

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 55

in triflate elimination to yield 20 (Scheme 5) which is difficult to separate from the product, throughout the remainder of the synthesis, even by chromatography. A conditions screen looking at the influence of bases, solvent, catalysts and temperature was conducted.

Scheme 5: Impurity formation during the sp2-sp3 Suzuki coupling

Pd(dppf)Cl2.CH2Cl2

was

found

to

be

a

superior

catalyst

to

the

original

bis(benzonitrile)palladium(II) chloride, Pd(PPh3)2Cl2, or the combination of palladium acetate and RuPhos. In 1984, Hayashi and co-workers21 had suggested that the high activity of Pd(dppf)Cl2 could be ascribed to the large P-Pd-P angle of the catalyst (99°). This, in turn, may explain the improved selectivity and reduction in side-products such as 20 caused by β-hydride elimination.22 Potassium phosphate remained the best base, but a lower number of equivalents was used (3 equiv instead of 6 equiv) to avoid overloading the reactor with solid material. Toxic triphenylarsine and solid silver oxide were eliminated, thereby considerably improving the workup. Other methyl donors such as trimethylboroxine or MeZnCl were explored with varying degrees of success, but did not afford improved conditions. Finally, a chromatographic step removed most impurities and controlled the level of by-product 20 down to 0.7%. As a result of

ACS Paragon Plus Environment

16

Page 17 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

these improved conditions, batches of C2-methyl product 18 could be produced in 44% yield and over two steps, on kg scale. Next, the nitro functionality was reduced with Zn and AcOH to provide aniline 19. Instead of pre-activating the zinc by washing it with dilute HCl, it was found more convenient to activate the zinc in situ by adding 5% water in the solvent (ethanol). Under these conditions, the reaction is rapid, even at low temperature (5 °C), and the exotherm can be controlled by the rate of addition of 18 to the reducing mixture. Amine 19 was isolated in 81% yield and 97% purity, (27% over 5 steps, 1 chromatography) and was used rapidly in the next steps due to its relatively low stability. Carbamate protection of Amine 19 Amine 19 occupies a key position in the synthesis as it is the starting material for two parallel branches (Figure 2): the top branch with a simple alloc protection (Scheme 6), and the bottom branch with the inclusion of the Val-Ala peptidic trigger. Both branches rejoin later in a dimerization step to form 39. Amine 19 was therefore split in two batches. A split factor was calculated based on subsequent yields and number of equivalents used in the dimerization. Initially, this split ratio was fixed at 0.39/0.61 top branch/bottom branch. However, as the yields improved and the number of equivalents in the dimerization step was optimized from 1.5 to 1.2 equiv, the split ratio changed to 0.58/0.42. It is likely that the ratio will be further optimized by subsequent campaigns improving the process robustness, and providing ever more reliable yields and equivalence factors. In any case, both branches follow the same chemistry, except for the carbamate formation with Alloc-Val-Ala-PAB-OH 24.

ACS Paragon Plus Environment

17

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 55

Scheme 6: Improved syntheses of carbamate protected PBD monomers 29 and 34

Reagents and conditions: (a) Allyl chloroformate, pyridine, DCM, -5 °C, 84%; (b) PTSA, THF, Water, 35 °C, 87%; (c) Oxalyl chloride, DMSO, triethylamine, DCM, -70 °C, 81%; (d) TBSOTf, 2,6-lutidine, DCM, 25 °C, 76%; (e) LiOAc, DMF, Water, 25 °C, 76%; (f) Triphosgene, 24, triethylamine, DCM, 25 °C, 69%; (g) PTSA, THF, Water, 25 °C, 68%; (h) Oxalyl chloride, DMSO, triethylamine, DCM, -60 °C; (i) TBS-OTf, 2,6-lutidine, DCM, -15 °C; (j) LiOAc, DMF, Water, 40 °C, 71% (3 steps and 1 chromatography).

ACS Paragon Plus Environment

18

Page 19 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

We have previously reported23 the synthesis of peptide trigger building block 24, following the studies of Dubowchik

24

and Jeffrey.25 In this work, we have optimized the synthesis to allow

production of highly pure material on 500 g scale (Scheme 7). Most notably, in step a, different bases were screened to avoid formation of double alloc, or dimerisation impurities. The combination of NaOH/ Na2CO3 showed an improved purity profile. In step d, prolonged slurrying with MTBE controlled the by-product level of EEDQ condensation (quinoline) to very low levels. Compound 24 was obtained at a purity of 98.9%, in 48% yield over 4 steps and did not require chromatography. The chiral purity was found to be 99.9% by chiral HPLC. Scheme 7: Improved synthesis of Alloc-Val-Ala-PAB-OH (24)

Reagents and conditions: (a) Allyl chloroformate, NaOH, Na2CO3, water, MTBE, 20 °C, 97%; (b) HOSu, DCC, DCM, 20 °C, 92%; (c) L-alanine, Na2CO3, THF, water, 25 °C, 79% (d) 4aminobenzyl alcohol, EEDQ, THF, 25 °C, 68%.

Protection of amine 19 with allyl chloroformate and pyridine was straightforward, even on kg scale, and did not require further improvements. However, the hydrolysis of primary TBS ether

ACS Paragon Plus Environment

19

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 55

25 originally relied on AcOH, in a mixture of water, THF and methanol. The removal of AcOH on scale during the work-up was difficult. Instead, a method relying on low number of equivalents of acid was investigated. Ultimately, 0.6 equiv of tosic acid (PTSA) in wet THF was used, thus considerably simplifying the work-up. Flash chromatography purification was performed here, in order to provide a high grade of material 26 going into the next reaction. Oxidation of the primary alcohol gives an aldehyde which spontaneously ring-closes to form the B-ring of the PBD system (step c, Scheme 6). The Swern reaction is well suited to this transformation26, and more so on manufacturing scale than on research scale. Indeed, the number of equivalents of oxalyl chloride had to be precisely controlled, or impurities such over-oxidized lactam and unreacted started material were observed. The best outcomes were obtained after a number of trial reactions, narrowing down on the optimum number of equivalents of oxalyl chloride. It goes without saying that the moisture content of the starting material was a key parameter, and Karl-Fisher determination helped inform the team when selecting a range of oxalyl chloride equivalences. Typically, a small excess was used to compensate for the remaining moisture. In this case, with a 0.23% water content (0.07 equiv water), 1.03 equiv of oxalyl chloride was used and the product 27 was found pure enough to be used as such in the next step (starting material / product / over-oxidation, 0.5 / 94.3 / 3.9). Protection of the secondary alcohol with TBS triflate and 2,6-lutidine did not require further optimization. Similarly, cleavage of the phenolic silyl ether with lithium acetate in wet DMF, proved to be a remarkably smooth, mild, high yielding and scalable method. A simple slurrying sequence in hexane and ethyl acetate provided key PBD monomer 29 with 99% purity and 34% yield over 5 steps, with a single chromatographic purification.

ACS Paragon Plus Environment

20

Page 21 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

The second PBD monomer 34 was synthesized in an almost identical sequence, apart from the peptide trigger introduction by isocyanate chemistry. The chloroformate of benzylic alcohol 24 could not be used, since these types of compounds tend to eliminate CO2 and form their chlorobenzyl analogues. On the other hand, aromatic aniline 19 is not nucleophilic enough to easily react with activated carbonates (although it must be noted that Smith and co-workers recently reported the clean condensation of an analogue of 19, at room temperature over a period of 6 days, with a pentafluorophenyl carbonate27). For these reasons, the research route relied on formation of an isocyanate intermediate by reaction of amine 19 with triphosgene (a solid, safer alternative to phosgene gas). Several attempts were made to substitute triphosgene with low toxicity reagents, but in these instances, were found inferior to the isocyanate and alcohol condensation. A key improvement to this reaction was the simple solvent switch from (hygroscopic) THF to DCM.28 Moisture levels had to be stringently controlled or urea sideproducts were observed as a result of isocyanate hydrolysis and self-condensation. Peptide 24 had a lower solubility in DCM than THF, but this was not found to be a limiting factor. These improved conditions allowed the number of equivalents of 24 to be lowered from 1.5 to 1.05, which in turn considerably simplified the work-up and chromatography. (An excess of 24 prevents silica gel chromatography by forming insoluble gel networks). Carbamate 30 was thus obtained in 69% yield. TBS deprotection with PTSA revealed primary alcohol 31 as above. Structure-activity relationship (SAR) of PBD species highlighted the importance to treat the synthetic intermediates with caution from this point onward. Unless forming part of a pro-drug strategy, ring-opened PBDs lacking an imine moiety (or equivalent carbinolamine and aldehydes) are relatively non-toxic29, and so are PBDs protected with biologically non-cleavable carbamates such as Alloc. For example, the IC50 of alloc protected phenol 29 could not be

ACS Paragon Plus Environment

21

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 55

measured (>10 µΜ, K562 CellTiter96 (MTS), 96 h incubation). But ring-closed intermediate 32 is protected by an enzymatically-cleavable dipeptide trigger, and could exert cytotoxicity in-vivo. A change of strategy was therefore applied. The synthesis was completed in facilities equipped to handle highly-potent compounds. Batch splitting and chromatography were permitted if necessary, and high performance preparative chromatography was considered as a purification option. Primary alcohol 31 was ring-closed by Swern oxidation as described for compound 26. Again, the oxalyl chloride equivalence was key, and in direct relation with the moisture content of the starting material. For example, if thorough water azeotroping with dry toluene was conducted, typical water levels would be at 0.06% w/w, and 1.03 equiv of oxalyl chloride were used. On the other hand, when azeotroping with toluene was not used, and the water content was measured at 0.2% w/w, 1.2 equiv of oxalyl chloride was used to fully consume the starting material and limit the formation of over-oxidized impurity 36 (Scheme 9). The use of lower temperature could not be considered to control over-oxidation due to lower solubility of the starting material in DCM. Because of the cytotoxic properties of these PBD intermediates, isolation in solid form was not a preferred option. Instead, the material was kept in solution (based on appropriate stability data in solution) and used directly in the next step. Protection of secondary alcohol 32 with TBS-OTf and 2,6-lutidine initially gave unacceptable levels of impurities. Mass analysis of the crude LC profile showed up to 13% of TBS-OTf mediated carbamate cleavage product 35, presumably in a reaction analogous to the BOC cleavage described by Sakaitani and Ofhune30 (Scheme 8).

Scheme 8: Postulated benzyl carbamate cleavage with TBS-OTf and 2,6-lutidine.

ACS Paragon Plus Environment

22

Page 23 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

A conditions screen looking at temperature and number of equivalents concluded that -15 °C was low enough to avoid this side reaction, whilst allowing the protection to proceed (albeit slowly). An excess of TBS-OTf (5 equiv) and 2,6-lutidine (6 equiv) was used to obtain full conversion in 20h. The solubilization of TBS-OTf in DCM proved to be useful for a better control of the addition. Again, this step was telescoped with the next one to avoid any solid isolation. The phenolic triisopropylsilyl ether was cleaved with lithium acetate in wet DMF as before, but it was found that the reaction kinetics could be improved at 40°C without any degradation. The extraction solvent was changed from ethyl acetate to Me-THF to avoid carrying traces of DMF in the organic phase, which would have negatively impacted on the subsequent normal phase chromatography. The removal of acid 37 (resulting from over-oxidation during the Swern reaction, Scheme 9) was efficient with additional NaHCO3 wash. Remarkably, this three steps sequence yielded pure phenol 34 after high pressure chromatography and precipitation in 71% yield. This key intermediate was isolated as a solid.

Scheme 9: Postulated conversion of over-oxidation impurity 36 to acid 37.

ACS Paragon Plus Environment

23

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 55

Next, the two phenolic monomers 29 and 34 were dimerized in two separate Williamson etherification steps. In both steps, acetone was replaced with MEK to allow higher operating temperatures, and a fine grade (typically ≤250 µm) of potassium carbonate was used to improve reaction kinetics. First, iodoalkane 38 was made by reacting phenol 29 with 5 equivalents of diiodopentane. This high number of equivalents not only drives the reaction kinetics forward, but also helps control the amount of homodimerization to below 10%. An aqueous work-up was implemented to remove the reaction salts. Iodoalkane 38 was isolated by high pressure chromatography and was stored in solution in ethyl acetate and used as such. This strategy presents some advantages on scale due to the oily nature of 38 which prevents straightforward transfers. In the second Williamson etherification, 38 and 34 were condensed under similar conditions. An important factor here is the excess of iodoalkane 38 versus phenol 34. The discovery route employed 1.5 equiv of 38 to fully consume 34 and ensure straightforward chromatography. However, as was discussed above, this equivalence factor is taken into account 7 steps earlier to calculate the proportion of amine 19 to commit to each branch. When the yields of the lower branch were dramatically improved, the amount of iodoalkane 38 became unbalanced with the quantities of phenol 34 available. Fortunately, the dimerization was found to work cleanly with a reduced equivalence of 38 (1.2 equiv), which partially restored the balance

ACS Paragon Plus Environment

24

Page 25 of 55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

between the two branches of the route. A good conversion was obtained and crude dimer 39 was taken directly to the next step without any purification.

Scheme 10: Final stages. Synthesis of tesirine (SG3249).

ACS Paragon Plus Environment

25

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 55

Reagents and conditions: (a) Diiodopentane, K2CO3, MEK, 75 °C, 88%; (b) 34, K2CO3, MEK, 75 °C; (c) TBAF, AcOH, THF, 20 °C, 89% (2 steps); (d) Pd(PPh3)4, pyrrolidine, DCM, 20 °C; (e) Mal-(PEG)8-acid, EDCI, DCM/MeOH, 20 °C, 75% (2 steps).

We have previously described how unbuffered fluoride deprotection of the secondary TBS ethers in 39 caused partial racemization of a key stereocenter in C11a.9 This base driven mechanism can be prevented by buffering the mixture with a mild acid such as AcOH. Although this represented a successful solution to the problem, we quickly noticed that the reaction kinetics dropped in line with the pH – the reaction being unacceptably slow at acidic pH (< 4). This

ACS Paragon Plus Environment

26

Page 27 of 55

meant that a relatively narrow and optimum pH range (around 6.0) was controlled by the precise equivalence of acid and TBAF. To add to our difficulties, and to our surprise, the actual fluoride content of the TBAF solutions varied considerably between batches. These inconsistencies were observed on multiple occasions, in different countries. Figure 4: AcOH titration of a weak batch of TBAF

8.5 8 7.5

pH

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Organic Process Research & Development

7 6.5 6 5.5 3.9

4.1

4.3

4.5

4.7

4.9

5.1

5.3

5.5

TBAF equivalents

Note: Titration of 4 equiv of AcOH with a weak batch of TBAF in THF/water 80/20. For example, when the reaction would not reach completion during one of the synthetic campaigns, an acid-base titration (Figure 4) revealed that 5 “equiv” of this batch of TBAF were required to neutralize 4 equiv of AcOH. The ratio used during the experiment was 4/ “3.9” AcOH/TBAF which at this pH (5.5) prevented the reaction proceeding rapidly. The actual concentration of the TBAF solution was found to be 0.72M (by anion-exchange chromatography; see supporting information)31 instead of 1.0M, thus explaining these inconsistencies. During a subsequent campaign, the reaction proceeded very rapidly, and partial

ACS Paragon Plus Environment

27

Organic Process Research & Development 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 55

racemization was detected in the final step. This time, the concentration was calculated to be 1.09 M instead of 1.0 M. On addition, fluoride precipitation can occur in bottles stored at 2 to 3 °C. With such wide error margin existing between real concentrations, and stated concentrations, we introduced a strict quality control of the TBAF solutions, to determine their exact concentrations before proceeding with the reaction (see supporting information for titration method). In order to ensure the robustness of this critical step, a DoE study (supporting information) was carried out to evaluate the impact of the main parameters (number of equivalents of TBAF per TBS, and ratio TBAF/AcOH) on the conversion and diastereoisomer content. The ratio TBAF/AcOH was the most important parameter to control the diastereoisomer content. A low variability was found in the defined range 0.5-0.8, where the model was stable. The amount of TBAF had a low impact on the diastereoisomer content. On the other hand, the model was found to be linear for the conversion response. The amount of TBAF can be increased to improve the kinetics of the reaction, while keeping a constant ratio of TBAF/AcOH. This DoE screen revealed that the reaction kinetics could be improved by adding further TBAF/AcOH, and that a margin of safety could be conserved by carrying out the reaction with a slight excess of AcOH. The final conditions were 1.5 equiv of TBAF per TBS, and 0.63 equiv of TBAF per AcOH. In these conditions, the reaction proceeded quite slowly (48 h), but guaranteed a low level of racemization (